Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Implement --train-align-safety Using DPO from TRL #2

Open
mrs83 opened this issue Nov 27, 2024 · 0 comments
Open

Implement --train-align-safety Using DPO from TRL #2

mrs83 opened this issue Nov 27, 2024 · 0 comments

Comments

@mrs83
Copy link
Owner

mrs83 commented Nov 27, 2024

Add support for the --train-align-safety feature to fine-tune the model using Direct Preference Optimization (DPO) from the TRL library.

This feature will align the model's responses to prioritize safe and preferred outputs (chosen) over unsafe or less-preferred ones (rejected).

The dataset should be structured with prompt, chosen, and rejected fields.

The aligned model should be robust against adversarial and unsafe prompts while maintaining high-quality responses.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

1 participant